EU AI Act transparency rules apply from 2 August 2026

The transparency obligations in Article 50 of the EU AI Act apply from 2 August 2026, two years after the Act entered into force on 1 August 2024. According to GDPR Local and Forbes, the obligations cover chatbot disclosure, machine-readable marking of AI-generated content, emotion recognition notices and deepfake labels.

What Article 50 requires

GDPR Local's explainer sets out four obligations. Under Article 50(1), people must be told when they're interacting with an AI system such as a chatbot. Article 50(2) requires AI-generated content to be marked in a machine-readable way, and Article 50(3) requires notice when emotion recognition is used. Article 50(4) covers disclosure of deepfakes and of AI-generated text published on matters of public interest.

What changed under the AI Omnibus

The AI Omnibus provisional agreement, reached in May 2026, gives generative AI systems already on the market before 2 August 2026 until 2 December 2026 to meet the machine-readable marking requirement. Forbes reports that the other Article 50 obligations apply from 2 August 2026 with no transition period. The extension only covers systems that were already on the market before that date, and it only applies to marking.

What it means for engineering teams

Most of this work lands in the product backlog. A chatbot disclosure needs a UI component, agreed copy and a decision on where and when it appears in the flow. Machine-readable marking means writing metadata or watermarks into generated content at the point of creation, then checking that your export, storage and delivery pipeline doesn't strip them. Emotion recognition notices and deepfake labels need their own interface states and the logic to trigger them.

Your team will probably also want logging that records when a disclosure was shown and which outputs were marked, so you can demonstrate what the product does if a regulator or customer asks. Legal teams can interpret the text, but the engineers write the code that displays a notice or attaches a marker. Teams with generative features already live have until 2 December 2026 for marking, while chatbot disclosure and the other obligations apply from 2 August 2026.

What it means when you hire engineers

If your roadmap now includes disclosure UI, content marking and audit logging, the engineers doing that work need to understand how your AI features produce output and where that output travels. Engineers who build with AI tools every day tend to know where generated content enters a system and where metadata gets lost along the way. Every Miyagami engineer completes two sixty-minute assessments in a real codebase, the first with AI tools off and the second with the candidate's own agents on, and you receive both scorecards. Shortlists arrive within five working days of a brief.

This item summarises public reporting from GDPR Local and Forbes and isn't legal advice. Check your own obligations with qualified counsel.

You can read more about how we vet engineers before they join your team.